Information Performance Tradeoffs in Control

INTRODUCTION Over the past few decades the fields of communication and control have been developed extensively to support the emerging needs of the society such as higher speed internet, better telephone communication networks, wireless access to devices over Internet of Things (IoT), reliable long distance satellite communication, home automation and many other related fields. While communication theory is mainly concerned with the transmission of information from one point to another reliably it does not deal with what is done with the information once it is transmitted. Control theory, on the other hand, mainly deals with using the information it receives as feedback in order to achieve better performing systems, which are not prone to disturbances, noises and other process variations. Although there is substantial research on control and information theory, the research on networked control systems (control systems with communication between its components) is still in a nascent stage. Networked control systems are increasingly finding wide applications but they also pose newer challenges which are neither addressed in control theory nor in information theory. The components of a networked control system are physically distributed with communication links between the plant, controller, observer and/or the actuators. These links connecting the different components are often noisy and pose various other constraints (See [1] for details). Hence, the control design and analysis needs to be done keeping in mind these information bottlenecks. In control system design, it is usually desired to minimize a performance index. Because of the information bottlenecks, there exists a tradeoff between the best achievable performance and the observation accuracy (it is intuitive that worse the observation accuracy, worse would be the performance of the system). We dealt with two important settings of networked control systems in this project: 1. Noiseless rate-limited communication of data between observer and controller 2. Communication over Additive White Gaussian Noise (AWGN) channel between the observer and the controller. 2 2 BACKGROUND